crystals.ai: Accelerating Materials Science with AI-Powered Tools and Datasets

crystals.ai is a cutting-edge platform developed by the Materials Virtual Lab at the University of California, San Diego, dedicated to advancing materials science through artificial intelligence. It offers a comprehensive suite of open-source tools, datasets, and software frameworks tailored for researchers and developers in the field. With powerful libraries like M3GNet, MEGNet, CGCNN, AENet, and GATGNN, users can build and apply graph neural networks to predict crystal properties, stability, and material behaviors with high accuracy. The platform supports a wide range of applications—from molecular property prediction using QM9 to advanced interatomic potential modeling with SNAP and specialized datasets for garnets and perovskites. Designed for scientific innovation, crystals.ai empowers researchers to explore new materials faster, enabling breakthroughs in energy, electronics, and sustainable materials development.

Key Features:

• M3GNet: Universal interatomic potential based on 3-body graph networks
• MEGNet: Graph network library for materials science applications
• MAterials Machine Learning (maml): Framework for machine learning in materials research
• Matminer: Advanced data mining tool for materials data analysis
• CGCNN: Crystal Graph Convolutional Neural Network for crystal property prediction
• AENet: Artificial neural network-based atomic interaction potential construction
• GATGNN: Graph Attention Network for enhanced inorganic materials property prediction
• MP 2019.4.1 Dataset: ~133,000 crystal structures with formation energies
• MP 2018.6.1 Dataset: ~60,000 crystal structures with formation energies, band gaps, and elastic constants
• QM9 Molecules Dataset: ~134,000 molecular graphs for quantum chemistry analysis
• SNAP Datasets: For developing Spectral Neighbor Analysis Potentials across elements and alloys
• Garnet and Perovskite Datasets: Pre-trained model files for stability prediction
• Open-source software frameworks with active community support
• Integration with blockchain and decentralized data systems (via UCSD’s research infrastructure)

Pricing: crystals.ai is offered as a free, open-access platform with no cost to use its tools, datasets, and frameworks. It supports academic and research use without financial barriers, making it ideal for collaborative innovation.

Conclusion: crystals.ai stands as a vital resource in materials science AI, combining powerful libraries, rich datasets, and open access to drive discovery and accelerate the development of next-generation materials.